qdrant-mcp-server

概述

Qdrant MCP Server 提供本地优先架构的全面语义搜索功能，支持多种嵌入提供商，包括 Ollama、OpenAI、Cohere 和 Voyage AI。服务器具有先进的代码向量化功能，支持 AST 感知的分块、Git 历史索引，以及结合语义和关键字搜索的混合搜索。可以通过 stdio 传输方式在本地部署，或作为带有身份验证支持的 HTTP 服务器远程部署。

试试问 AI

装完之后，这里有 6 个你可以让 AI 做的事：

• create_collection

  Create collection with specified distance metric (Cosine/Euclid/Dot)

  试试:Create a collection called 'research_papers' with cosine distance试试:Create a new document collection named 'notes'
• add_documents

  Add documents with automatic embedding (supports string/number IDs, metadata)

  试试:Add a document about machine learning to my collection试试:Insert multiple research papers with metadata tags
• semantic_search

  Natural language search with optional metadata filtering

  试试:Search for documents about quantum computing试试:Find all articles mentioning 'blockchain' from 2023
• hybrid_search

  Hybrid search combining semantic and keyword (BM25) search with RRF

  试试:Find documents about 'neural networks' using both semantic and keyword search试试:Search for 'API documentation' with hybrid approach
• index_codebase

  Index a codebase for semantic code search with AST-aware chunking

  试试:Index my TypeScript project for code search试试:Index the backend codebase with AST-aware chunking
• search_code

  Search indexed codebase using natural language queries

  试试:Search for authentication middleware in the codebase试试:Find database connection patterns in the project
• index_git_history

  Index git commit history for semantic search over past changes and fixes

  试试:Index git history of this repository for semantic search试试:Index the commit history to search for bug fixes
• search_git_history

  Search indexed git history using natural language queries

  试试:Search git history for authentication implementation试试:Find commits related to database schema changes
• contextual_search

  Combined code + git history search with file-commit correlations

  试试:Search for authentication implementation across code and git history试试:Find how user management evolved over time in code and commits
• federated_search

  Search across multiple repositories with Reciprocal Rank Fusion (RRF) ranking

  试试:Search for payment processing patterns across all repositories试试:Find API implementations across multiple codebases
• reindex_changes

  Incrementally re-index only changed files (detects added/modified/deleted)

  试试:Update the index with recent code changes试试:Re-index only the modified files in the project
• delete_documents

  Delete specific documents by ID

  试试:Remove document with ID 'doc123' from the collection试试:Delete outdated research papers from the database

可对比工具

安装

# 克隆并安装
git clone https://github.com/mhalder/qdrant-mcp-server.git
cd qdrant-mcp-server
npm install

# 启动服务
podman compose up -d   # 使用 Podman
docker compose up -d   # 使用 Docker

# 拉取嵌入模型
podman exec ollama ollama pull nomic-embed-text

# 构建
npm run build

{
  "mcpServers": {
    "qdrant": {
      "type": "stdio",
      "command": "node",
      "args": ["/path/to/qdrant-mcp-server/build/index.js"]
    }
  }
}

FAQ

支持哪些嵌入提供商？

该服务器支持 Ollama（默认）、OpenAI、Cohere 和 Voyage AI 作为嵌入提供商。

我可以使用 Qdrant Cloud 吗？

可以，您可以使用 API 密钥身份验证连接到 Qdrant Cloud 或受保护的实例。

qdrant-mcp-server 对比